Learning Objectives: Stress hyperglycemia (SH) of critical illness is multifaceted; it involves elevated stress hormones, insulin resistance (IR), and insulin deficiency (ID). The effect of insulin drip adjustments on glucose can be mathematically modeled for various IR and ID combinations, providing insight into the glucoseinsulin dynamics found in SH. Methods: A glucose-insulin model was used to simulate two groups of 20 virtual patients: those receiving, or not receiving, a continuous nutritional source; SH was created by doubling the gluconeogenesis rate, varying IR from borderline to high, and changing insulin secretion from a normal to severely deficient state. This produced pre-insulin glucose oscillations with maximum (Gmax) values between 160-170 mg/dl with minimum (Gmin) values between 80-115 mg/dl. Next, an insulin infusion was titrated to produce two tight glycemic control (TGC) goals with Gmax values of 110 and 125 mg/dl; the insulin was then abruptly stopped. After the glucose oscillations returned to pre-insulin levels, the insulin infusions were restarted at the same TGC rates. The effects on glucose oscillations were studied. Results: Abruptly stopping an insulin infusion caused significant temporary rebound hyperglycemia (RH) when TGC 110 mg/dl infusions were the starting point, provided IR was high (largest increase 20 mg/dl) without dependence on ID. Abruptly restarting an insulin infusion resulted in Gmin values that were temporarily lower than those in the steady state insulin infusions (largest decrease 20 mg/dl); this was most pronounced when the TGC 110 mg/dl rates were used with severe ID. With no nutritional source, RH was not produced; however, mild hypoglycemia occurred when the insulin infusions were restarted when IR was high. Conclusions: Glucose-insulin dynamics are nonlinearly affected by IR, ID, and the nutritional source; each of which may change over time. It is unlikely that a traditional insulin drip protocol, based on a single simple algorithm, could optimally handle such complexity. Mathematical glucose-insulin models may prove the most adaptive in the critical care setting.